Restores the VTT/SRT, Fixes Comments, and etc

examples/prerecorded/file/main.go

@@ -62,4 +62,18 @@ func main() {
 		os.Exit(1)
 	}
 	log.Printf("\n\nResult:\n%s\n\n", prettyJson)
+
+	vtt, err := res.ToWebVTT()
+	if err != nil {
+		log.Printf("ToWebVTT failed. Err: %v\n", err)
+		os.Exit(1)
+	}
+	log.Printf("\n\n\nVTT:\n%s\n\n\n", vtt)
+
+	srt, err := res.ToSRT()
+	if err != nil {
+		log.Printf("ToSRT failed. Err: %v\n", err)
+		os.Exit(1)
+	}
+	log.Printf("\n\n\nSRT:\n%s\n\n\n", srt)
 }

pkg/api/live/v1/interfaces/types.go

@@ -4,9 +4,9 @@
 
 package interfaces
 
-/*
-Shared defintions for the Deepgram API
-*/
+/***********************************/
+// shared/common structs
+/***********************************/
 // Word is a single word in a transcript
 type Word struct {
 	Confidence     float64 `json:"confidence,omitempty"`
@@ -42,9 +42,9 @@ type Metadata struct {
 	RequestID string    `json:"request_id,omitempty"`
 }
 
-/*
-Results from Live Transcription
-*/
+/***********************************/
+// Results from Live Transcription
+/***********************************/
 // MessageResponse is the response from a live transcription
 type MessageResponse struct {
 	Channel      Channel  `json:"channel,omitempty"`

pkg/api/manage/v1/manage.go

@@ -2,6 +2,9 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
+/*
+This package contains the code for the Keys APIs in the Deepgram Manage API
+*/
 package manage
 
 import (

pkg/api/prerecorded/v1/interfaces/prerecorded.go

@@ -0,0 +1,66 @@
+// // Copyright 2023 Deepgram SDK contributors. All Rights Reserved.
+// // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
+// // SPDX-License-Identifier: MIT
+
+/*
+This package provides the types for the Deepgram PreRecorded API.
+*/
+package interfaces
+
+import (
+	"encoding/json"
+	"errors"
+	"fmt"
+	"net/http"
+	"strings"
+)
+
+// ToWebVTT implements output for VTT
+func (resp *PreRecordedResponse) ToWebVTT() (string, error) {
+	if resp.Results.Utterances == nil {
+		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
+	}
+
+	vtt := "WEBVTT\n\n"
+	vtt += "NOTE\nTranscription provided by Deepgram\nRequest ID: " + resp.Metadata.RequestID + "\nCreated: " + resp.Metadata.Created + "\n\n"
+
+	for i, utterance := range resp.Results.Utterances {
+		utterance := utterance
+		start := SecondsToTimestamp(utterance.Start)
+		end := SecondsToTimestamp(utterance.End)
+		vtt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
+	}
+	return vtt, nil
+}
+
+// ToSRT implements output for SRT
+func (resp *PreRecordedResponse) ToSRT() (string, error) {
+	if resp.Results.Utterances == nil {
+		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
+	}
+
+	srt := ""
+
+	for i, utterance := range resp.Results.Utterances {
+		utterance := utterance
+		start := SecondsToTimestamp(utterance.Start)
+		end := SecondsToTimestamp(utterance.End)
+		end = strings.ReplaceAll(end, ".", ",")
+		srt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
+
+	}
+	return srt, nil
+}
+
+func SecondsToTimestamp(seconds float64) string {
+	hours := int(seconds / 3600)
+	minutes := int((seconds - float64(hours*3600)) / 60)
+	seconds = seconds - float64(hours*3600) - float64(minutes*60)
+	return fmt.Sprintf("%02d:%02d:%02.3f", hours, minutes, seconds)
+}
+
+func GetJson(resp *http.Response, target interface{}) error {
+	defer resp.Body.Close()
+
+	return json.NewDecoder(resp.Body).Decode(target)
+}

pkg/api/prerecorded/v1/interfaces/types.go

@@ -2,9 +2,14 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
+/*
+This package provides the types for the Deepgram PreRecorded API.
+*/
 package interfaces
 
+/***********************************/
 // share/common structs
+/***********************************/
 type Metadata struct {
 	TransactionKey string   `json:"transaction_key,omitempty"`
 	RequestID      string   `json:"request_id,omitempty"`
@@ -134,7 +139,9 @@ type SummaryV2 struct {
 }
 type Summary SummaryV2 // internal reference to old name
 
-// Response
+/***********************************/
+// response/result structs
+/***********************************/
 type PreRecordedResponse struct {
 	RequestID string   `json:"request_id,omitempty"` // for ?callback=...
 	Metadata  Metadata `json:"metadata,omitempty"`

pkg/api/prerecorded/v1/prerecorded.go

@@ -121,52 +121,3 @@ func (c *PrerecordedClient) FromURL(ctx context.Context, url string, options int
 
 	return &resp, nil
 }
-
-// func (resp *PreRecordedResponse) ToWebVTT() (string, error) {
-// 	if resp.Results.Utterances == nil {
-// 		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
-// 	}
-
-// 	vtt := "WEBVTT\n\n"
-
-// 	vtt += "NOTE\nTranscription provided by Deepgram\nRequest ID: " + resp.Metadata.RequestId + "\nCreated: " + resp.Metadata.Created + "\n\n"
-
-// 	for i, utterance := range resp.Results.Utterances {
-// 		utterance := utterance
-// 		start := SecondsToTimestamp(utterance.Start)
-// 		end := SecondsToTimestamp(utterance.End)
-// 		vtt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
-// 	}
-// 	return vtt, nil
-// }
-
-// func (resp *PreRecordedResponse) ToSRT() (string, error) {
-// 	if resp.Results.Utterances == nil {
-// 		return "", errors.New("this function requires a transcript that was generated with the utterances feature")
-// 	}
-
-// 	srt := ""
-
-// 	for i, utterance := range resp.Results.Utterances {
-// 		utterance := utterance
-// 		start := SecondsToTimestamp(utterance.Start)
-// 		end := SecondsToTimestamp(utterance.End)
-// 		end = strings.ReplaceAll(end, ".", ",")
-// 		srt += fmt.Sprintf("%d\n%s --> %s\n%s\n\n", i+1, start, end, utterance.Transcript)
-
-// 	}
-// 	return srt, nil
-// }
-
-// func SecondsToTimestamp(seconds float64) string {
-// 	hours := int(seconds / 3600)
-// 	minutes := int((seconds - float64(hours*3600)) / 60)
-// 	seconds = seconds - float64(hours*3600) - float64(minutes*60)
-// 	return fmt.Sprintf("%02d:%02d:%02.3f", hours, minutes, seconds)
-// }
-
-// func GetJson(resp *http.Response, target interface{}) error {
-// 	defer resp.Body.Close()
-
-// 	return json.NewDecoder(resp.Body).Decode(target)
-// }

pkg/api/version/constants.go

@@ -2,6 +2,7 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
+// This package handles the versioning in the API for live/streaming transcription
 package version
 
 import "errors"

pkg/api/version/manage-version.go

@@ -2,9 +2,7 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
-/*
-This package handles the versioning in the API both async and streaming
-*/
+// This package handles the versioning in the API for live/streaming transcription
 package version
 
 import (

pkg/audio/microphone/constants.go

@@ -2,7 +2,7 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
-// Implementation microphones using portaudio
+// Implementation microphone using portaudio
 package microphone
 
 const (

pkg/audio/microphone/microphone.go

@@ -2,7 +2,7 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
-// Implementation microphones using portaudio
+// Implementation microphone using portaudio
 package microphone
 
 import (
@@ -19,15 +19,15 @@ import (
 func Initialize() {
 	err := portaudio.Initialize()
 	if err != nil {
-		klog.V(1).Printf("portaudio.Initialize failed. Err: %v\n", err)
+		klog.V(1).Infof("portaudio.Initialize failed. Err: %v\n", err)
 	}
 }
 
 // Teardown cleans up the library
 func Teardown() {
 	err := portaudio.Terminate()
 	if err != nil {
-		klog.V(1).Printf("portaudio.Terminate failed. Err: %v\n", err)
+		klog.V(1).Infof("portaudio.Terminate failed. Err: %v\n", err)
 	}
 }
 
@@ -97,7 +97,6 @@ func (m *Microphone) Stream(w io.Writer) error {
 			}
 
 			byteCount, err := w.Write(m.int16ToLittleEndianByte(m.intBuf))
-			// byteCount, err := w.Write(m.int16ToLittleEndianByte(m.intBuf))
 			if err != nil {
 				klog.V(1).Infof("w.Write failed. Err: %v\n", err)
 				return err

pkg/audio/microphone/types.go

@@ -2,7 +2,7 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
-// Implementation microphones using portaudio
+// Implementation microphone using portaudio
 package microphone
 
 import (

pkg/client/interfaces/types-prerecorded.go

@@ -3,7 +3,7 @@
 // SPDX-License-Identifier: MIT
 
 /*
-This package contains the interface to manage the prerecorded interfaces for the Deepgram API
+This package contains the interface to manage the prerecorded and live/stream interfaces for the Deepgram API
 */
 package interfaces
 

pkg/client/interfaces/types-stream.go

@@ -3,7 +3,7 @@
 // SPDX-License-Identifier: MIT
 
 /*
-This package contains the interface to manage the live/streaming interfaces for the Deepgram API
+This package contains the interface to manage the prerecorded and live/stream interfaces for the Deepgram API
 */
 package interfaces
 

pkg/client/interfaces/utils.go

@@ -2,6 +2,9 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
+/*
+This package contains the interface to manage the prerecorded and live/stream interfaces for the Deepgram API
+*/
 package interfaces
 
 import (

pkg/client/live/constants.go

@@ -2,6 +2,9 @@
 // Use of this source code is governed by a MIT license that can be found in the LICENSE file.
 // SPDX-License-Identifier: MIT
 
+/*
+This package provides the live/streaming client implementation for the Deepgram API
+*/
 package live
 
 import (

pkg/client/live/init.go

@@ -3,7 +3,7 @@
 // SPDX-License-Identifier: MIT
 
 /*
-This package contains the initialization code for the Deepgram Go SDK
+This package provides the live/streaming client implementation for the Deepgram API
 */
 package live